J. Phys. Chem.

A 2009, 113, 4397–4402 4397

Surface Raman Spectroscopy of Chemistry at the Tris(8-hydroxyquinoline) aluminum/Ca

Interface†

Robert J. Davis and Jeanne E. Pemberton*

Department of Chemistry, UniVersity of Arizona, 1306 E. UniVersity BouleVard, Tucson, Arizona 85721
ReceiVed: December 15, 2008

Surface Raman spectroscopy in ultrahigh vacuum is used to interrogate interfaces formed between tris(8-
hydroxyquinoline) aluminum (Alq3) and vapor-deposited Ca. Vapor deposition of Ca onto Alq3 films results
in the appearance of new vibrational modes consistent with the formation of a Ca-Alq3 adduct in which Ca
is bound through the O of Alq3. In addition, the graphitic carbon is observed to form with the deposition of
Ca onto Alq3 films. The frequency and relative peak height ratios of graphitic D-band and G-band observed
are consistent with more disordered sp3-type carbon forming upon partial decomposition of the organic film.

Introduction ficiency including the formation of new species through

chemical reaction, interface dipole formation, and diffusion of
Tris-8-hydroxyquinoline aluminum (Alq3) has become a materials at the interface. On the basis of the electron affinity
common electroluminescent and electron transport material in of Alq3, 2.9 eV,26 and the ionization potentials of Al, Ca, and
organic light emitting diodes (OLEDs) due to its superb stability Mg of 6.0, 6.1, and 7.6 eV, respectively,27 it is expected that
and luminescence properties.1,2 Due to its prevalence and an electron will be transferred from the metal atom to Alq3 at
importance in device function, the interface of Alq3 with the interface of these materials.
cathodic metals has been examined extensively by UPS/XPS,3-10
Theoretical studies of the interaction of Al with Alq3 have
vibrational spectroscopy,11-15 and theory16-20 in an effort to
reported stable metal-Alq3 adducts in which the metal atom is
relate interfacial chemical and electronic properties to device
bound to Alq3 through the O atom of the 8-hydroxyquinoline
performance. Metals used as cathodes are commonly selected
(8-HQ) ligand, while studies of Mg with Alq3 have reported
to have a low work function in order to minimize any barrier
Mg to be bound to both the O and C atoms of the 8-HQ
to electron injection from the cathode to the electron transport
ligand.16-20 Shifts in the XPS core levels for the N 1s and O 1s
layer. Precise tuning of this charge injection barrier is essential
core levels reported in studies of Mg-Alq3 and Al-Alq3
to producing high efficiency devices. If the charge transfer rate
interfaces support the formation of such complexes.5-10 Previous
between the metallic cathode and the organic transport layer is
studies of these interfaces by Raman spectroscopy in this
too low, overall device efficiency will be reduced, while if the
laboratory have reported the formation of new ν(M-O) modes
charge transfer rate is significantly higher than the electron
and shifts in the ν(C-O), ν(C-N), and νring modes of pristine
transport rate of the bulk organic layer, electrons will accumulate
Alq3 upon deposition of Mg and Al onto Alq3 thin films.14,15
at the metal-organic interface also leading to a reduction in
These spectral changes are also consistent with formation of
device performance. Thus, a precise understanding of the
metal-organic complexes in which Mg and Al are bound to
chemical and electronic interfacial properties is essential to
Alq3 through the O atom of 8-HQ.
designing high efficiency devices.
DFT calculations of the interaction of Ca atoms with Alq3
Ca has been effectively used as a cathode in high efficiency
by Curioni and Andreoni have also found stable Ca-Alq3
polymer-based devices based on PPV,21 MEH-PPV,22 and
structures in which Ca is bound to one O atom of 8-HQ in mer-
poly(9,9-dioctylfluorene)23 while Ca is less commonly used in
Alq3 and to three O atoms in fac-Alq3.16 Interestingly, the
conjunction with Alq3 in devices. Ca, with a work function of
binding energies for the calculated Ca-Alq3 complexes (11 kcal/
2.8 eV,3,10 would appear to be an ideal cathode material for use
mol for Ca-mer-Alq3, 23 kcal/mol for Ca-fac-Alq3) are
in conjunction with Alq3 whose LUMO level has been placed
substantially lower than those reported for the Al-Alq3
by optical measurements at 2.9 eV,24 leading to a minimal
complexes (37 kcal/mol for Al-mer-Alq3, 51 kcal/mol for
theoretical barrier to charge injection. A study of efficiency for
Al-fac-Alq3), indicating weaker binding for the Ca-Alq3
devices incorporating Alq3 with several low work function
system.16 UPS and XPS studies of Ca-Alq3 have reported that
metals as cathodes, however, has reported that Ca/Alq3 does deposition of Ca onto Alq3 films leads to the decomposition of
not perform as well as might be expected.25 The highest the organic in a staged reaction, attributed to the formation of
electroluminescent efficiency for interfaces using Alq3 was the Alq3 radical anion at Ca coverages e4 Å and decomposition
observed by Stossel et al.25 for Mg, having a work function of of Alq3 at higher coverages as evidenced by peak splitting of
3.7 eV, with a 10-15% dropoff in relative efficiency noted for the N and O 1s core levels and changes in elemental
lower work function metals, such as Ca (2.8 eV) and Li (2.95 stoichemiometry.3,4 Interestingly, partial decomposition of Alq3
eV). This trend in device performance is often attributed to other to graphitic carbon has been reported for Al-Alq3 and Mg-Alq3
interfacial properties that can also affect charge transfer ef- in the Raman spectroscopy of these systems.14,15 Vibrational
† modes associated with formation of amorphous, nanocrystalline
Part of the “George C. Schatz Festschrift”.
* To whom correspondence should be addressed. E-mail: pembertn@ graphitic carbon are present for these systems and attributed to
u.arizona.edu. Phone: 520-621-8245. Fax: 520-621-8248. a reaction of excess deposited metal with the metal-Alq3
10.1021/jp8110698 CCC: $40.75  2009 American Chemical Society
Published on Web 02/25/2009
4398 J. Phys. Chem. A, Vol. 113, No. 16, 2009 Davis and Pemberton

complexes.14,15 A similar decomposition reaction could explain

the shift of the C 1s to lower binding energy reported in XPS
studies of Ca-Alq3.3,4
The high reactivity of these metal-organic interfaces further
complicates efforts to characterize them. The potential for oxide
formation by low work function metals such as Ca has been
shown to degrade device performance and requires the use of
ultrahigh vacuum (UHV) to prevent extrinsic interface reactions
due to metal oxide formation.28,29 Previous studies of the
interaction of Al-trans-stilbene, Ag-Alq3, Al-Alq3, and
Mg-Alq3 by Raman spectroscopy in UHV have demonstrated
the technique’s high specificity and sensitivity to changes in
structure and bonding at metal-organic interfaces.14,15,30 In this
study, Raman spectroscopy is used to assess the reactivity of
Ca deposited onto Alq3 thin films as a function of metal coverage
in UHV.

Experimental Section
Ultrahigh Vacuum Chamber. Organic and metallic thin
films were prepared and characterized in a custom-built ultrahigh
vacuum (UHV) chamber.14,15,30 Thin films were deposited in a
sample preparation chamber separated by a 6 in. gate valve and
analysis chamber. A 300 L s-1 turbomolecular pump, an ion
pump, and two titanium sublimation pumps maintain a base
pressure of 2 × 10-10 torr in the analysis chamber; working
pressures were seldom above 10-9 torr. The ion and titanium Figure 1. Peak fitting of Raman spectra of 150 Å Alq3 films after Ca
sublimation pumps are liquid nitrogen-cooled to remove con- deposition of (a) 0, (b) 5, (c) 10, (d) 15, and (e) 20 Å mass thickness. Raw
densable gases (primarily H2O). Raman spectrum (black solid lines), overall fit envelope (black dashed
Chemically polished Ag substrates were mounted on a 1.5 m lines), Alq3 modes (red dashed lines), Ca-Alq3 reaction product modes
Advanced Research Systems DE-204B closed-cycle He-cooled (green dashed lines), and graphitic carbon modes (blue dashed lines).
cryogenic translatable arm that could be cooled to a base
operating temperature of 30 K. This arm can be translated
respect to the surface normal. Collected radiation was collimated
between the sample preparation and analysis chambers at
and focused onto the front entrance slit of a single monochro-
temperature. Temperature was controlled in these studies using
mator (Spex 270M) with a 1200 gr/mm grating and set for a 5
a Cryo-Con Model 34 cryogenic temperature controller using
cm-1 spectral bandpass. A holographic SuperNotch Plus filter
a type K, Chromel-AuFe thermocouple at the bottom of the
(Kaiser Optical Systems) was used to reduce Rayleigh scattering.
coldfinger within close proximity of the stub.
A 1340 × 400 pixel, thinned, back-illuminated CCD (Roper
Materials. 8-Hydroxyquinoline, aluminum salt (99.995%),
Scientific model 400-EB) held at -95 °C was used for detection.
was obtained from Aldrich and was degassed at 250 °C at 10-9
Typical integration times varied from 10 to 30 min per
torr for 6 h prior to deposition. Calcium crystalline dendritic
acquisition. Spectra were calibrated using Ar+ emission lines.
pieces (99.98%) were obtained from Alpha Aesar and degassed
Peak fitting of Raman spectra was accomplished with Grams/
at 400 K at ∼1 × 10-9 torr for 1 h prior to remove any metal
32 using 100% Gaussian peaks. Peaks at frequencies associated
oxide. A 1.25 cm diameter Ag (111) disk (99.999%, Alfa Aesar)
with intense Alq3 modes were fit using peak frequencies ((5
was used as the underlying substrate for all experiments.
cm-1) and relative intensities ((10%) determined by Halls and
Procedures. The procedure for obtaining chemically polished
Aroca33 with full-width-at-half-maximum (fwhm) values of 15
Ag surfaces has been outlined previously.31,32 Alq3 films were
cm-1 ((10 cm-1). Modes associated with graphitic carbon were
prepared on Ag substrates held at 300 K, similar to the
fit using five broad bands using peak frequencies ((10 cm-1),
conditions used in the fabrication of OLED devices, and are
relative intensities ((50%), and fwhm ((50%) described by
spectroscopically consistent with the formation of amorphous
Ferrari and Robertson.36 All additional spectral features were
or R-Alq3.33-35 Ca was deposited after degassing onto the
fit with broad bands that were allowed to vary in frequency by
Alq3-Ag sample held at ∼60 K from Al2O3-coated W/Ta
(10 cm-1, in fwhm by (10 cm-1, and intensity by (50%.
crucibles to ensure a sticking coefficient of 1 for the deposited
metals and to prevent the desorption of any volatile reaction
Results and Discussion
products. Film deposition was monitored differentially using a
multifilm deposition monitor (Maxtek TM-400) and a gold- Ca has a low atomic ionization potential that supports electron
coated quartz sensor crystal (Maxtek SC-101). The sensor was transfer to the conjugated 8-HQ ring of Alq3, and UPS/XPS
held at room temperature for Alq3 deposition but cooled with measurements have demonstrated a strong Ca interaction with
liquid N2 during metal deposition; reported values of coverage the organic layer.3,4 Figure 1 shows Raman spectra at 90 K of
represent mass thicknesses of the metal assuming bulk density. a pristine 150 Å Alq3 film on Ag before and after deposition of
Raman spectra were acquired by moving the sample stub into Ca layers ranging in mass thickness from 5 to 20 Å. With the
the glass cell at the end of the analysis chamber. Radiation of deposition of 5 Å Ca in Figure 1b, the Alq3 modes at 1595,
514.5 nm with 20 mW power at the sample was provided by a 1394, 543, 525, and 504 cm-1 exhibit a 30-40% intensity
Coherent Innova 350C Ar+ laser. Light was incident on the decrease compared to those from the pristine Alq3 thin film.
sample at 60°, and scattered photons were collected at 30° with Broad modes also appear for 5 Å Ca on Alq3 from 1100 to
Surface Raman Spectroscopy
Figure 2. Normalized Raman peak intensity as function of metal
coverage for Ca deposited onto Alq3 for [ring deformation + ν(Al-O)]
at 525 cm-1 (black square), graphitic carbon δring at 573 cm-1 (red circle),
graphitic carbon D-band at 1315 cm-1 (green up triangle), [νring +
δ(C-H)] at 1394 cm-1 (blue down triangle), graphitic carbon G-band
at 1515 cm-1 (light blue cross), and νring at 1595 cm-1 (yellow right
triangle). All modes normalized to intensities of respective bands in
pristine Alq3.
1600 cm-1 in the region associated with the formation of
graphitic carbon.
In order to more readily observe the differing spectral
contribution from Alq3, graphitic carbon, and any other reactive
products, peak fitting was employed. Peak fitting of the region
from 1100 to 1600 cm-1 using 100% Gaussian bands reveals
two intense, broad bands at 1505 and 1265 cm-1 whose breadth
and frequency are consistent with the presence of graphitic
carbon.36 In addition, a broad mode present at 577 cm-1
corresponds to a weak out-of-plane vibrational mode of
amorphous graphite, previously only reported in sp3-rich graph-
ite.37 The peak at 1515 cm-1 is attributed to the G-band of
graphitic carbon, with the known low-frequency asymmetry of
the G-band fit using an additional Gaussian band at 1435 cm-1.
The peak at 1315 cm-1 is attributed to the D-band of graphitic
carbon. As with the G-band, the known low-frequency asym-
metry of the D-band is fit using an additional broad band at
1200 cm-1. Similar modes have been previously observed in
Raman spectra for the Mg-Alq3 and Al-Alq3 systems and are
attributed to a decomposition reaction resulting from the
interaction of excess metal atoms with metal-Alq3 complexes
formed at the interface. Interestingly, the intensity ratio of the
D-band to G-band is higher than that reported for Mg-Alq3
(∼0.7) or Al-Alq3 (∼0.5),14,15 indicating that the graphitic
carbon formed at the Ca-Alq3 interface is more disordered than
for the other metal-Alq3 systems. The ratio of peak intensities
of the fit bands for the D-band to the G-band is ∼1.1, indicating
the presence of largely amorphous carbon with some sp3
character.36 The amount of graphitic carbon also increases
monotonically from 5 to 20 Å Ca in Figure 2 while the ratio of
the D:G band remains relatively constant. The lack of a D′-
band at 1605-1620 cm-1 was observed in both the Mg-Alq3
and Al-Alq3 systems and indicates small grain graphitic carbon
with little clustering of graphite into larger sheets.36 The low
frequency asymmetry of both the D-band and G-band graphitic
modes, as shown in the fit peaks at 1200 and 1435 cm-1,
respectively, is also more pronounced for Ca-Alq3 than that
reported for other metal-Alq3 systems. A similar increase is
the low frequency asymmetry of the D- and G-bands of graphitic
carbon has been noted after sputtering of highly ordered
polycrystalline graphite (HOPG) with N+ ions.38 The authors
of this study attributed the increased asymmetry of the D- and
G-bands to ion implantation into the graphite film, leading to
J. Phys. Chem. A, Vol. 113, No. 16, 2009 4399
more discontinuities in the crystalline structure. A similar
argument may explain the highly disordered graphitic carbon
observed in the Ca-Alq3 films in which excess Ca, as well as
other decomposition products of Alq3, hinder the formation of
graphite with long-range order.
Upon deposition of 5 Å of Ca onto Alq3, new modes also
appear at 1472, 1440, 1306, 790, 618, and 577 cm-1 that are
not consistent with modes for Alq3, although these modes appear
in frequency regions consistent with structures closely related
to 8-hydroxyquinoline and/or with Alq3 modes that are not
normally Raman-active.14,33 The new modes present at 1472,
1440, and 1306 cm-1 with Ca coverage of 5 Å are in the
frequency region are associated with Alq3 ring stretching and
ring deformation modes. The decrease in frequency of these
modes relative to those for pristine Alq3 at 1595 and 1394 cm-1
is explained by binding of Ca to Alq3, resulting a lowering of
ring electron density, leading to a decrease in the force constant.
The product mode at 1306 cm-1 is additionally assigned to a
ν(C-O) mode shifted from the mode at 1335 cm-1 for pristine
Alq3.14,33 A similar shift in the ν(C-O) mode has been reported
for both Al-Alq3 and Mg-Alq3,14,15 respectively, and is attributed
to binding of the metal.
The absence of any observable ν(Ca-C) modes from 300 to
500 cm-1 in Figure 1b suggests that Ca does not bond directly
to the ring. Thus, it can be concluded that the shifts in ring
breathing modes must be an indirect electronic effect due to
removal of ring electron density by interaction of Alq3 with the
deposited Ca. New modes also appear at 790 cm-1 with 5 Å
Ca deposited and at 841 cm-1 at 10 Å Ca deposited in the region
associated with δ(C-H)wag and δring modes, respectively. These
modes are shifted to lower frequency from the δ(C-H)wag at
862 cm-1 and δring at 804 cm-1,33 again as a consequence of a
reduction in electron density in the π-system of Alq3 upon
binding of Ca.
The mode at 618 cm-1 in Figure 1b is assigned as a ν(Ca-O)
resulting from binding of Ca to the O atom of the 8-HQ in
Alq3. Binding of the inserted Ca in this position is congruous
with the theoretical structures determined by Curioni and
Andreoni16 in which the lowest energy structure has the inserted
Ca bound through the O atom of a mer-Alq3 molecule as
depicted in Figure 3. The authors additionally found a structure
for the Ca-(fac-Alq3) adduct in which the Ca was proposed to
bind to three O atoms; this structure is more stable than the
Ca-(mer-Alq3) structure with 1-coordinate binding to O. These
authors also theorized that mer-Alq3, the more stable Alq3 isomer
that is presumed to be dominant in thin films, may convert to
fac-Alq3 upon interaction with deposited Al.16 However, the
multitude of spectral changes in the spectral region of the Raman
spectrum associated with Al-N and Al-O bonding upon Ca
deposition prohibits definitive confirmation of the conversion
from mer- to fac-Alq3, and either Ca-(fac-Alq3) or Ca-(mer-
Alq3) or both are possible products. It should be noted that the
binding energies calculated for Ca-(fac-Alq3) and Ca-(mer-
Alq3) complexes of 23 and 11 kcal/mol, respectively, are
significantly lower than those calculated for the Al-(fac-Alq3)
and Al-(mer-Alq3) complexes of 51 and 39.5 kcal/mol,
respectively.16 These values indicate that the Ca-Alq3 complex
is less stable than the Al-Alq3 adduct and may explain the
strong tendency toward decomposition to graphitic carbon noted
for this system.
Choong et al.,4 in their UPS study of this interface, observed
for the deposition of 0.5 Å Ca onto Alq3 the formation of a gap
state along with a concomitant loss of Alq3 spectral features.
The gap state indicates a significant change in chemical and
4400 J. Phys. Chem. A, Vol. 113, No. 16, 2009 Davis and Pemberton

TABLE 1: Raman Peak Frequencies and Assignments for a

Pristine Alq3 Film and after Deposition of 20 Å of Ca
peak frequency (cm-1)
literature 150 Å 20 Å Ca
Alq3a Alq3 on Alq3 mode assignmentb
1595 1595 1595 νring
1515 graphite G-band
1472 νring
1471 1440 νring
1435 graphite G-band
1424 1424 νring + δ(C-H)
1394 1394 1394 νring + δ(C-H)
1386 δ(C-H) + νring
1315 graphite D-band
1335 1306 νring + ν(C-O) + δ(C-H)
1200 graphite D-band
1060 1060 1060 ν(C-O) + δ(C-H)
862 841 δ(C-H)wag
804 790 δring
757 757 757 νring
647 647 647 ν(Al-O) + ν(Al-N) + δring
618 ν(Ca-O)
577 577 δring + δ(Al-O-C)
573 graphite δring
541 543 543 ν(Al-O) + ν(Al-N) + δring
525 525 525 δring + ν(Al-O)
504 504 504 δring
470 470 CH torsion
a
From ref 23. b Assignments for Alq3 from ref 23 and those for
graphite from refs 24 and 28.

lack of changes to the ν(Al-O), ν(Al-N), and δ(Al-O-C)

Figure 3. Proposed Ca-Alq3 adduct structures in which Ca is bonded
to Alq3 through ligand O atoms for mer-isomer (top) of Alq3 and fac-
isomer (bottom) of Alq3. C, H, N, O, and Al atoms of Alq3 shown in
gray, white, blue, red, and green, respectively; vapor-deposited Ca atom
shown in yellow.
electronic properties, even at low Ca coverages, as would be
expected for the formation of a Ca-Alq3 adduct. XPS measure-
ments by the same authors demonstrated a shift in the C, N, O
1s core levels to higher binding energies by ∼1 eV, that was
attributed to band bending resulting from a change in the
interface dipole with 4 Å Ca deposited onto Alq3 films. At higher
coverages of Ca (15-30 Å), an additional weak intensity N 1s
core level peak is reported at ∼0.5 eV lower binding energy
from the N 1s of the Alq3 film.4 A significantly more intense O
1s core level peak is observed at higher Ca coverages, with a
shift by 1 eV to lower binding energy compared the O 1s of
Alq3. The presence of this new band supports an increase in
electron density upon the proposed binding of Ca with the O
atom of 8-HQ and is confirmed by the shift in ν(C-O) to lower
frequency and the appearance a new ν(Ca-O) in Figure 1b-e.
Broadening of the νring at 757 cm-1 also occurs for 5-20 Å
Ca deposited. A similar broadening has been reported for this
mode in the Raman spectra of Al-Alq3 and Mg-Alq3.14,15 The
modes upon exposure to Ca that were reported for Mg-Alq315
and Al-Alq314 may be due to a weaker interaction between Ca
and Alq3 than the interaction reported for Mg and Al. This weak
interaction results in the magnitudes of shifts in Alq3 electron
density upon bonding with the metal that are not as significant
in Ca/Alq3.
The spectral changes that are observed for the Ca-Alq3
system are summarized in Table 1. These changes suggest that
a complex interface is formed that is composed of highly
disordered graphitic carbon as well as a Ca-Alq3 adduct that
has a structure similar to that proposed in Figure 3, in which
the deposited Ca is bound through the O atom of the 8-HQ
ligand. It is also important to note that the intensities of the
Ca-Alq3 product modes relative to those of pristine Alq3 are
lower than that reported for either Al-Alq3 or Mg-Alq3.14,15
The smaller degree of Ca-Alq3 adduct formation may result
from the instability of the Ca-Alq3 complex or from an inability
of Ca to penetrate into the Alq3 film. The atomic radius of Ca
(1.80 Å) is considerable larger than that of either Mg or Al
(1.50 and 1.25 Å).27 Figure 4 shows a Chem 3D rendition of a
Ca inserted between two molecules in R-Alq3. The large size
of the Ca atom places it in close contact with the neighboring
Alq3 ligand, separated by only 1.5-2.0 Å, and may limit
penetration of Ca into the Alq3 film, thereby retarding the
formation of the Ca-Alq3 complex.
The temperature used for the deposition of Ca metal films in
this study of 60 K was chosen to ensure a high sticking
coefficient of Ca on the Alq3 film and retention of any volatile
reactive species. This temperature is less than that typically used
in the construction of organic electronic devices from these
materials. Although this temperature is not believed to signifi-
cantly alter the structures of the Alq3 film or products formed
between Ca and Alq3, cooling of the samples may result in
reduced diffusion of the deposited metal, limiting the depth of
Surface Raman Spectroscopy
Figure 4. Proposed structure for Ca inserted into R-crystal phase of
mer-Alq3. C, N, O, and Al atoms of Alq3 shown in gray, blue, red, and
green, respectively; vapor-deposited Ca atom shown in yellow.
the reaction with Alq3 and creating a narrower metal-organic
interfacial region than may be formed in real devices produced
at room temperature.
Conclusions
Raman spectra for Alq3 films exhibit significant spectral
changes after deposition of 5-20 Å Ca. The loss in intensity
of all major Alq3 modes coupled with the appearance of new
modes supports the partial reactive conversion of Alq3 to a new
product in which Ca is bound to Alq3 through the O atom of
8-HQ as evidenced by the appearance of a new ν(Ca-O) mode
at 618 cm-1. The shifts in νring and ν(C-O) further support
interaction of the deposited Ca with an O atom of Alq3 and a
subsequent shift in electron density in the π-system of the ligand.
The relative intensities of the Ca-Alq3 product modes observed
here are significantly lower than those reported previously for
both the Al-Alq3 and Mg-Alq3 products, indicating, as
proposed by Curioni and Andreoni, that the Ca-Alq3 complex
may be significantly less stable than the other metal-Alq3
complexes.16
The appearance and growth of graphitic carbon modes at
1515, 1435, 1315, 1200, and 573 cm-1 indicate degradation of
Alq3 to graphitic carbon after the deposition of Ca. The
frequencies and relative intensity ratio of the D-band and G-band
modes of graphitic carbon indicate that highly disordered
graphite with a large sp3 character is formed as a result of partial
decomposition of the organic film.36 This graphitic carbon is
considerably more disordered and contains more sp3 character
than the graphitic carbon previously reported for both Mg-Alq3
and Al-Alq3, illustrating that important chemical differences
are present at the reactive interfaces of these seemingly similar
metals with Alq3.
The presence of even a small amount of graphitic carbon at
the Ca-Alq3 interface has profound implications for interfacial
electronic structure and conductivity in functioning electronic
devices. The graphitization reaction reported in this study at
90 K is expected to be retained at room temperature, com-
mensurate with the conditions used for device fabrication,
although diffusion may limit the extent of the reaction and
breadth of the interfacial region at 90 K. The formation of a
Ca-Alq3 complex at the interface may also have several effects
on charge injection. The Ca-Alq3 complex likely lies at an
J. Phys. Chem. A, Vol. 113, No. 16, 2009 4401
energy level between the bulk Alq3 film and the metal Fermi
level, as observed by the formation of a gap state in UPS
examination of the interface.5 Such an interfacial state may act
as an intermediate hopping state between the metal and Alq3,
thus lowering the thermal energy required for injection.39 The
presence of Ca-Alq3 complexes as well as decomposition
products also likely creates disorder within the Alq3 film at the
interface, further broadening the interfacial region and the energy
distribution of interfacial states. This broadening may also lower
the effective thermal energy required for electron injection.
Lastly, the presence of decomposition products at the Ca/
Alq3 interface may further broaden the energy distribution of
the interfacial states toward lower energies. Notably, the
graphitic carbon formed at Ca/Alq3 has a higher sp3 character
than that observed at either Al/Alq3 or Mg/Alq3. This more
disordered carbon contains a higher degree of defect sites which
may serve as traps for injected electrons, lowering the efficiency
of charge transfer across the interface. Conversely, the presence
of additional lower energy states from the disordered graphitic
carbon may aid in electron injection by serving as hopping sites
between the metal and Alq3 levels. The electroluminescent
efficiencies reported by Stossel et al.25 for the Ca/Alq3 interface
suggest that the former behavior may dominate in these systems,
or in other words, the high number of defect sites in the graphitic
carbon trap injected electrons. Despite similar electron affinities
for Mg and Ca, Raman spectra of these metal-organic interfaces
clearly show that the structures of both the metal-Alq3 adducts
and the graphitic carbon formed in these systems are substan-
tially different.
Acknowledgment. The authors gratefully acknowledge sup-
port of this research by the National Science Foundation.
References and Notes
(1) Tang, C. W.; Vanslyke, S. A. Appl. Phys. Lett. 1987, 51, 913.
(2) Tang, C. W.; Van Slyke, S. A.; Chen, C. H. J. Appl. Phys. 1989,
65, 3616.
(3) Park, Y.; Choong, V. E.; Hsiesh, B. R.; Tang, C. W.; Wehrmeister,
T.; Mullen, K.; Gao, Y. J. Vac. Sci. Technol., A 1997, 15, 2574.
(4) Choong, V. E.; Mason, M. G.; Tang, C. W.; Gao, Y. Appl. Phys.
Lett. 1998, 72, 2689.
(5) Rajagopal, A.; Kahn, A. J. Appl. Phys. 1998, 84, 355.
(6) Le, Q. T.; Yan, L.; Choong, V.-E.; Forsythe, E. W.; Mason, M. G.;
Tang, C. W.; Gao, Y. Synth. Met. 1999, 102, 1014.
(7) Le, Q. T.; Yan, L.; Gao, Y.; Mason, M. G.; Giesen, D. J.; Tang,
C. W. J. Appl. Phys. 2000, 87, 375.
(8) Shen, C.; Hill, I. G.; Kahn, A.; Schwartz, J. J. Am. Chem. Soc.
2000, 122, 5391.
(9) Shen, C.; Kahn, A.; Schwartz, J. J. Appl. Phy. 2001, 89, 449.
(10) Yan, L.; Mason, M. G.; Tang, C. W.; Gao, Y. L. Appl. Surf. Sci.
2001, 175, 412.
(11) Sakurai, Y.; Salvan, G.; Hosoi, Y.; Ishii, H.; Ouchi, Y.; Seki, K.;
Kampen, T. U.; Zahn, D. R. T. Appl. Surf. Sci. 2002, 190, 382.
(12) Sakurai, Y.; Hosoi, Y.; Ishii, H.; Ouchi, Y.; Salvan, G.; Kobitski,
A.; Kampen, T. U.; Zahn, D. R. T.; Seki, K. J. Appl. Phys. 2004, 96, 5534.
(13) Sakurai, Y.; Yokoyama, T.; Hosoi, Y.; Ishii, H.; Ouchi, Y.; Salvan,
G.; Kobitski, A.; Kampen, T. U.; Zahn, D. R. T.; Seki, K. Synth. Met. 2005,
154, 161.
(14) Davis, R. J.; Pemberton, J. E. J. Phys. Chem. C 2008, 112, 4364.
(15) Davis, R. J.; Pemberton, J. E. J. Am. Chem. Soc., submitted.
(16) Curioni, A.; Andreoni, W. J. Am. Chem. Soc. 1999, 121, 8216.
(17) Zhang, R. Q.; Lu, W. C.; Lee, C. S.; Hung, L. S.; Lee, S. T. J. Chem.
Phys. 2002, 116, 8827.
(18) Meloni, S.; Palma, A.; Schwartz, J.; Kahn, A.; Car, R. J. Am. Chem.
Soc. 2003, 125, 7808.
(19) Meloni, S.; Palma, A.; Kahn, A.; Schwartz, J.; Car, R. J. Appl.
Phys. 2005, 98, 23707.
(20) Takeuchi, K.; Yanagisawa, S.; Morikawa, Y. Sci. Technol. AdV.
Mater. 2007, 8, 191.
(21) Blom, P. W. M.; deJong, M. J. M.; Vleggaar, J. J. M. Appl. Phys.
Lett. 1996, 68, 3308.
(22) Parker, I. D. J. Appl. Phys. 1994, 75, 1656.
4402 J. Phys. Chem. A, Vol. 113, No. 16, 2009
(23) Brown, T. M.; Friend, R. H.; Millard, I. S.; Lacey, D. J.;
Burroughes, J. H.; Cacialli, F. Appl. Phys. Lett. 2001, 79, 174.
(24) Burrows, P. W.; Forrest, S. R. Appl. Phys. Lett. 1994, 64, 2285.
(25) Stossel, M.; Staudigel, J.; Steuber, F.; Simmerer, J.; Winnacker,
A. Appl. Phys. A: Mater. Sci. Process. 1999, 68, 387.
(26) Schlaf, R.; Parkinson, B. A.; Lee, P. A.; Nebesny, K. W.;
Armstrong, N. R. Appl. Phys. Lett. 1998, 73, 1026.
(27) CRC Handbook of Chemistry and Physics, 64th ed.; Weast, R. C.,
Ed.; CRC Press: Boca Raton, FL, 1984.
(28) Shen, C. F.; Hill, I. G.; Kahn, A. AdV. Mater. 1999, 11, 1523.
(29) Broms, P.; Birgerson, J.; Salaneck, W. R. Synth. Met. 1997, 88,
255.
(30) Hawkridge, A. M.; Pemberton, J. E. J. Am. Chem. Soc. 2003, 125,
624.
Davis and Pemberton
(31) Smolinski, S.; Zelenay, P.; Sobkowski, J. J. Electroanal. Chem.
1998, 442, 41.
(32) Tiani, D. J.; Pemberton, J. E. Langmuir 2003, 19, 6422.
(33) Halls, M. D.; Aroca, R. Can. J. Chem. 1998, 76, 1730.
(34) Brinkmann, M.; Gadret, G.; Muccini, M.; Taliani, C.; Masciocchi,
N.; Sironi, A. J. Am. Chem. Soc. 2000, 122, 5147.
(35) Curry, R. J.; Gillin, W. P.; Clarkson, J.; Batchelder, D. N. J. Appl.
Phys. 2002, 92, 1902.
(36) Ferrari, A. C.; Robertson, J. Phys. ReV. B 2000, 61, 14095.
(37) Tamor, M. A.; Vassell, W. C. J. Appl. Phys. 1994, 76, 3823.
(38) Watanabe, H.; Takahashi, K.; Iwaki, M. Nucl. Instrum. Methods
Phys. Res., Sect. B 2007, 257, 549.
(39) Baldo, M. A.; Forrest, S. R. Phys. ReV. B 2001, 64, 85201.
JP8110698